TGFβ signaling is required during human and chick Neural Crest formation

Abstract

Neural crest (NC) cells are multipotent cells unique to vertebrates that arise early in development, at the edge of the neural plate, and subsequently undergo an epithelial to mesenchymal transition, migrate throughout the body, and differentiate into many different derivatives, contributing to the formation of many organs and systems. NC induction research from multiple modeling organisms has identified critical roles for a few signaling pathways, including Wnt, BMP, FGF, Notch/ Delta, Indian Hedgehog, and Endothelin signaling (Prasad et al., 2019). Given the limitations of human embryo studies, pluripotent stem cell models of human NC formation have provided a resourceful alternative (Lee et al., 2007). Intriguingly, while TGFβ inhibition had not been identified as a signaling requirement for NC formation in any in vivo model organism, several pluripotent stem cell (PSC) models of human NC induction rely on TGFβ inhibition (Chambers et al., 2009). To address this issue, we evaluate the role of TGFβ in NC formation using our human (hNC) model that depends on WNT signaling and does not require TGFβ inhibition (Leung et al., 2016, Gomez et al., 2019). We report that under our model, TGFβ signaling is required, and that moderate levels of TGFβ and pSMAD2 levels are necessary for optimal NC formation (with negative effects seen upon strong activation or inhibition). Moreover, we demonstrate that PSC cultured in mTeSR1 immediately prior to hNC induction instead required TGFβ modulation in addition to WNT signaling activation to render hNC. Using the chick embryo as an in vivo vertebrate model, we further provide evidence of expression and requirement of relevant TGFβ signaling components during NC formation. This study identifies an important role for TGFβ signaling in early NC development, opening the door for novel players as effectors mediating the multiple signals integrated during early neural crest development.

Graphical Abstract

Introduction

Neural Crest (NC) cells are a migratory cell population first described by Wilhelm His in 1868 (His, 1868), and whose intriguing biology was further elucidated by Julia Plat (Dupont 2021) in recognition of their ectomesenchymal potential. Studies in various vertebrate model organisms, from fish, frogs, and chicks to rodents and rabbits, have addressed their formation through tissue interactions and the signaling pathways, orchestrating their progression from progenitors to NC to their final terminal derivatives (Stuhlmiller 2012; Betters 2018; Mendez-Maldonado 2020; Prasad et al., 2019). In the late 1990s and early 2000s, with the advent of molecular tools, NC induction research focused on the signaling molecules responsible for NC formation and pinpointed critical roles for Wnt, BMP, FGF, Notch/Delta, Indian Hedgehog, and Endothelin signaling (Stuhlmiller 2012; Mendez-Maldonado 2020; Prasad et al., 2019). This information provided considerable gains that, until recently, excluded human biology, chiefly due to embryo inaccessibility, particularly at the relevant early stages.

Only a handful of studies have addressed embryonic human neural crest cells. Bondurand et al. reported a focused study on Sox10 expression during embryonic, fetal, and adult human tissues, yet the description of early NC development (4 and 6 weeks) was limited. O’Rahilly and Müller provide a great histological account of morphological and developmental events during NC formation in vertebrate embryos (O’Rahilly, R., & Müller, F. 2007). Betters et al. provided a more detailed expression profile for a battery of NC transcription factors Msx1/2, Pax3, Pax7, Sox9, Sox10, and AP-2, in addition to the cell surface proteins, human natural killer-1 (HNK-1) and p75 neurotrophin receptor (p75NTR), on a small collection of human embryos focusing on the 4^th-6^th weeks of embryonic development (from CS 12 to 18) broadly confirming that human NCs express markers found in model organisms, as expected. Other human NC studies analyzed the expression of NC markers at later developmental stages, focusing on NC derivatives (Tucker et al., 1984; Bondurand et al., 1998; Gershon et al., 2005; Benko et al., 2009) or analyzed mRNA expression in cell lines derived from human NCs (Thomas et al., 2008).

However, in 2005, work pioneered by Ronald Goldstein’s group opened the door to human models based on pluripotent embryonic stem cells (PSCs/ESCs) that produced human NC cells in vitro (Pomp et al., 2005). Initially, timely culture conditions requiring sorting and complex culture media led to more efficient, shorter, and tractable models (Lee 2007; Bajpai et al., 2010; Jiang et al.,2009; Hotta 2009; Menendez 2011). Amongst these efforts, our group developed a human model based on low-density cultures of hESCs by WNT induction and ROCK inhibition (CHIR/ROCKi) that generates human cranial (anterior) NC in just 5 days that expresses all the markers expected of NC and can be differentiated into expected NC terminal derivatives (Leung et al., 2016; Gomez 2019; Prasad et al., 2020; Charney et al., 2023).

Many other reported human models of NC formation based on PSCs include TGFβ inhibition in their culture conditions, apparently inherited from protocols aiming to generate neural derivatives (Lee 2007; Chambers 2009). Additionally, other groups reported a critical role for WNT signaling during mesoderm formation from hESCs/iPSCs (Funa 2015; Kreuser et al., 2020) and reported that under their conditions, TGFβ inhibition leads to NC formation (Funa 2015). Intriguingly, we previously reported that TGFβ inhibition under our culture conditions, CHIR/ROCKi at low density, had a negative effect and reduced NC marker expression (Leung et al., 2016). Thus, whether TGFβ promotes or prevents NC formation remains an unanswered question. Here, we further explored the impact of TGFβ signaling in our culture conditions and found that moderate levels of TGFβ signaling are required during human NC formation. We demonstrate that in these culture conditions, TGFβ-related phosphoSMAD2 is clearly detected and that TGFβ inhibition depletes phosphoSMAD2, which blocks NC induction. We also found that, in addition to CHIR/ROCKi, pre-cultured hESCs before initiation of NC differentiation require TGFβ inhibition dependent on pre-culture conditions. Then, to address whether TGFβ signaling is required in an amniote embryo, we electroporated dominant negative SMAD2/3 into chick embryos. Importantly, we found that blocking TGFβ signaling in gastrulating embryos prevents NC formation.

Materials and Methods

Neural Crest Differentiation

Anterior cranial neural crest cell induction was performed on the human embryonic stem cell (hESC) line H1 (WA01) obtained from WiCell Research Institute, Inc. (Madison, WI, USA), as described (Gomez et al., 2019, Leung 2016), and referred to in the Supplementary Table 1 as the “G-C Lab basic culture conditions”. hESCs were maintained in mTeSR1 (Stem Cell Technologies) at 37°C in 5% CO2, 5% O2, and passaged regularly with Versene (Thermo Fisher Scientific). For hNC differentiation, hESCs were collected at ~80% to 90% confluency. Cells were rinsed in Ca+ and Mg+ free-PBS (Thermo Scientific, Cat. No. 14190144), then dissociated in Accutase (Stem Cell Technologies). Accutase was quenched with DMEM/F12 (Invitrogen, Cat. No. 17504–044) containing 10μM Rock Inhibitor, Y-27632 (Tocris, Cat. No. 1254). The remaining cell pellet was resuspended in Neural Crest Induction media (DMEM-F12, 3% BSA (Sigma), 2% Glutamax (Stem Cell Technologies). The single-cell suspension was seeded at a density of 20,000 cells/cm² (unless noted otherwise) onto plates pre-coated with Matrigel. For normal inductions, cells were kept in Neural Crest Induction media with 10μM Rock Inhibitor, Y-27632, and 3μm CHIR99021 (Tocris, Cat. No. 4423) for the first two days, and maintained in induction media (without CHIR and Y-27632) from day 3 until day 5. Induction media was changed daily until the day of collection for further analysis. 10μM SB431542 (Tocris, Cat. No. 1614), 10μM SB505124 (Tocris, Cat. No. 3263), human recombinant TGF-β (Stem Cell Technologies, Cat. No. 78067), or equivalent volumes of DMSO were added on days indicated in the manuscript.

Immunofluorescence (IF)

Cells were fixed in 4% paraformaldehyde (PFA), rinsed in PBS, then permeabilized with 0.4% Triton X100, blocked with 4% fetal bovine serum, and incubated with primary antibodies [Mouse anti-SOX10 IgM: Santa Cruz (H-2),sc-271163; Mouse anti-PAX7 IgG1: DSHB, AB_528428; Goat anti-TBXT: R&D Systems AF2085; Mouse anti-GSC Abcam Ab58352; Rabbit anti-snai2 Cell-Signaling Technologies 9585] overnight at 4°C. Species-specific and Immunoglobulin-specific Alexa Fluor conjugated secondary antibodies (Invitrogen) were used at a 1:2000 dilution, washed, then nuclei were stained with DAPI (10μg/ml). Chicken embryos were collected at an appropriate developmental stage and fixed in 4% paraformaldehyde overnight at 4°C, and whole-mount immunofluorescence was performed as described (Prasad et al., 2020), with primary antibodies: Mouse anti-PAX7 antibody (DSHB, AB_528428); Rabbit anti-GFP (Millipore-Sigma AB3080).

Western Blots

Cells were lysed in RIPA lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 0.5% sodium deoxycholate, 1% NP40, 0.1%) with 1 mM DTT, 1X protease inhibitor, and 1X phosphatase inhibitor cocktail (Sigma-Aldrich). Protein lysate concentrations were determined with a Bradford protein assay (Bio-Rad), and 10 μg of each lysate was boiled in 4X SDS dye, then loaded on 10% SDS-PAGE gels for immunoblotting on Polyvinylidene fluoride (PVDF) membranes. Membranes were blocked in 4% BSA in 1X TBS and probed with rabbit anti-pSMAD2 1:1000 (138D4, Cell Signaling Technologies), rabbit anti-SMAD2 1:1000 (D43B4, Cell Signaling Technologies), or β-actin 1:10,000 (A5316, Sigma-Aldrich). Primary antibodies were detected with goat anti-rabbit IgG-HRP (31462, Thermo Scientific) or goat anti-mouse IgG-HRP (31432, Thermo Scientific), both at 1:10,000. Blots were detected with Immobilon Chemiluminescent HRP substrate (P90720, MilliporeSigma) and exposed on autoradiography film (1968-3057, USA Scientific). Scanned images were quantified on ImageJ software.

Cell Counts

Immunofluorescence images taken on NIKON eclipse 80i were converted to nd2 format and counted on NIS-Elements AR Analysis software (version 4.6). Nuclei were counted by the clustered method at 5px/spot, at a typical diameter of 10μm, with variable contrast. Positively immunostained nuclei were taken as a ratio to total nuclear cells, as detected by DAPI stain.

Microscopy

Images were captured on a Nikon Eclipse 80i microscope using a Spot SE camera and software or on an inverted Nikon Eclipse Ti microscope using NIS Elements software. All images were compiled and adjusted in Adobe Photoshop CS5. Statistical analysis of cell counts and graphs was performed on GraphPad Prism9.

Whole-mount in situ hybridization

mRNA expression in whole-mount chicken embryos was detected by standard methods as described (Charney et al., 2023). In situ probes were amplified by PCR from a pooled library containing cDNA from three different stages: EGXII, HH4, and HH10. A T7 promoter sequence including overhangs, shown as lowercase base pairs, was added in the reverse primer for improved transcription (aagcttTAATACGACTCACTATAGGGAGA) of DIG-labeled probes. Probes were generated by in vitro transcription from column-purified PCR amplicons. The following primers were used. Note: Underlined sequences are complementary to the target probe sequence.Smad2:[FWD:AGAGGAGAGGTTGGTGTGCTA]-[RVS:aagcttTAATACGACTCACTATAGGGAGATCCCACTGATCTATCGTGTTTGG](40 2bp).Smad3:[FWD:AGAGTGGAATTGGCTGAACCC]-[RVS:aagcttTAATACGACTCACTATAGGGAGACTTGCATGTGCTTTACACGCT](450bp).Tgfbr1:[FWD:CAAACCAGCAATTGCCCACA]-[RVS:aagcttTAATACGACTCACTATAGGGAGAGCACAGAAAGGACCCAAAGC] (738 bp).

Plasmid cloning and electroporation

Plasmid subcloning:

Dominant negative Smad3 and Smad2 plasmid constructs were obtained from Addgene [#14958; pCMV5 Flag-Smad2 (1-456) and #14964; CS2 Flag-Smad3 (3A)] to disrupt TGF-β signaling in chick embryos. Plasmids were sub-cloned into the pCIG vector, which contains an internal ribosome entry site for independent expression of eGFP as a tracer, using XhoI, XmaI, and SmaI restriction enzymes by traditional restriction enzyme cloning methods. Mutations in relevant coding sequences were verified by Sanger sequencing.

Chick embryo electroporation:

Chick embryos were obtained from Sunstate Ranch (Sylmar, CA) and staged according to Hamburger and Hamilton (Hamburger and Hamilton, 1951). Embryos were injected at the developmental stage of HH4 with a mixture of pCIG-SMAD2-DM and pCIG-SMAD3-DN plasmid constructs at 1mg/mL per plasmid, or with pCIG-empty vector control at 2mg/mL. Plasmids were reconstituted in 2% sucrose and 0.01% Fast Green. Embryos were immediately electroporated after injection with 5 pulses of 50ms ON, 50ms OFF at 6.5V on a B.T.C. electroporation supply source. Electroporated embryos were further subjected to the EC cultured method as described (Chapman 2001).

TGFβ inhibition in vivo with SB-soaked bead implants.

Approximately 50–100 Affi-Gel Blue beads (Bio-Rad) were coated with 0, 1, or 10μM SB431542 (Tocris, Cat. No. 1614) suspended in Ringer’s solution. Beads were incubated for 1h at room temperature and subsequently kept at 4°C or on ice. Beads were rinsed briefly in Ringer’s solution before implantation, then implanted onto the ectoderm of stage HH3 chick embryos, in prospective anterior neural fold/neural tube territories. Embryos were further incubated under EC culture for 12-18 hours, until they reached HH stages 9-10, at which point they were imaged to identify the bead’s final position, fixed with 4% PFA, and subjected to IF.

Statistics

Statistics were analyzed on Prism (Version 9.5.1). Cut-off levels for significance were determined with a P value <0.05.

Results:

We previously reported that prolonged activin/TGFβ antagonism had a negative effect on our human NC induction (hNCi) model (Leung 2016). Given that other hNC models instead require activin/TGFβ antagonism (Funa et al., 2015, Fukuta et al., 2014, Lee 2007, Menendez et al., 2020), we further analyzed the contribution of TGFβ inhibition in our hNC model. Here, we began by testing whether brief TGFβ antagonism with the TGFβ Type 1 receptor (ALK-4, ALK-5, ALK-7) inhibitor SB431542 (SB), administered daily, would affect hNCi in our improved/streamlined 5-day model (Gomez 2019). Briefly, freshly dissociated hESCs were plated at 20,000 hESCs/cm² and treated with 3 μM CHIR and ROCKi or DMSO vehicle (CON) on days 0-2, then maintained in NC induction media for 3 additional days. Cells were harvested on day 5 and evaluated for mature NC markers SOX10 and PAX7 by immunofluorescence. Treatment with CHIR alone resulted in robust expression of both markers as expected, while the addition of SB on the first and second day, but not thereafter, dampened both SOX10 and PAX7 expression (Figure 1A). In addition, we observed significant downregulation of TFAP2A, PAX3, PAX7, SNAI2, SOX9, SOX10, and FOXD3 mRNA expression by transcriptional analysis (data not shown). Moreover, TGFβ inhibition on the first day with SB505124, a different TGFβ small molecule receptor antagonist, also prevented hNCi (Supplemental Figure 1A). These data indicate that TGFβ activity is necessary for NC induction during the first 2 days in our hNCi model.

Figure 1. TGFβ is required for or antagonistic to NC formation, depending on pre-culture conditions.

(A) Representative immunofluorescence images on day 5 after a 2-day pulse with DMSO (CON), 3μM CHIR (CHIR), or a 2-day pulse of CHIR and addition of 10μM SB431542, SB, on days indicated at daily intervals. SOX10 is Green; PAX7 is Red; DAPI is Blue. (n = 2). (B) Immunofluorescence images of hESC colonies 5 days after switching from mTeSR1 to our NC culture medium containing 3μM CHIR on days 0-2 without or with 10μM SB on 0-1 or 0-2 days. SOX10 is Green; PAX7 is Red; DAPI is Blue (n = 1). (C) Immunofluorescence images of hESC colonies treated with either DMSO (CON) or 3μM CHIR for 5 days. TBXT is Green; GSC is Cyan; DAPI is Blue. Images were taken at the edge of colonies within dashed boxes of insets in overlaid images (n = 1).

Once we confirmed that TGFβ signaling is necessary for our hNCi model and its inhibition blocks NC formation, we turned our attention to a few plausible differences in culture conditions that could explain the difference in TGFβ requirements between hNCi models, including cell density, pre-seeding prior to initiation of hNCi, and the inclusion of mTeSR1 during pre-seeding. The different conditions tested and corresponding outcomes are summarized in Supplementary Table 1.

Some hNCi models that depend on TGFβ inhibition rely on higher cell densities, and it is plausible that high-density cultures produce a different signaling landscape influencing lineage-specific differentiation. We previously reported that an initial seeding density of 20,000 (20K) cells/cm² was optimal for hNCi (Leung et al., 2016), so here we tested the effect of TGFβ inhibition at different cell seeding densities in our model. As expected, CHIR induced robust SOX10 expression, indicative of hNCs, in hESCs seeded at 20K cells/cm², but only a few SOX10+ cells in hESCs seeded at 10K cells/cm² and 120K cells/cm² (Supplemental Figure 1B). By contrast, CHIR-mediated induction of SOX10 was dramatically reduced by 1 day of TGFβ antagonism at both 10K cells/cm² and 20K cells/cm² seeding densities relative to their density-matched counterparts treated with CHIR only (Supplemental Figure 1B). Instead, only a minor reduction in SOX10+ cells is noted between cells treated with CHIR+SB relative to CHIR alone after seeding at 120K cells/cm². Yet, it is clear that neither lower nor higher cell density triggers a requirement of TGFβ inhibition, as SB addition did not produce more SOX10 expression in these conditions (Supplemental Figure 1B). Therefore, differences in the initial hESC density subject to WNT activation do not change the requirement for TGFβ in our hNCi model.

Interestingly, varying from our model, where freshly dissociated PSCs are exposed to NC induction conditions (WNT), hNCi elicited by WNT activation and TGFβ antagonism has been reported when applied to hESCs colonies, which have been maintained or pre-cultured with mTeSR1 (Funa 2015, Fukuta 2014), and WNT activation alone instead triggered mesoderm formation (Funa 2015). To explore if these conditions could alter the signaling requirements for NC and mesoderm, we challenged hESC colonies (not freshly dissociated cells) by switching them from mTeSR1 maintenance media to our hNC induction conditions. Under these conditions, no NC formation was apparent in either control or WNT activation alone, but instead, WNT activation with TGFβ inhibition for either 0-1 or 0-2 days clearly led to NC formation (Figure 1B). Furthermore, treatment of hES colonies with WNT alone did display expression of axial mesoderm markers Brachyury (TBXT) and Goosecoid (GSC) (Figure 1C). Thus, in agreement with previous reports, hESC colonies pre-grown in mTeSR1 differentiate toward mesoderm upon WNT activation, while NC formation requires additional TGFβ antagonism (Funa 2015).

Next, we designed experiments to distinguish between the possible signaling effects of mTeSR1 treatment and pre-culture (no dissociation) on hNCi. First, we tested whether adherent hESCs pre-seeded in our basal induction media (DMEM/F12, B27, BSA) could be induced to NCs by a 2-day pulse of CHIR treatment. To this end, dissociated hESCs were plated and either induced immediately with 2-day WNT activation or maintained in our induction media for 1 day prior to WNT activation. NC formation was evaluated 5 days after the initiation of WNT activation by the addition of CHIR (Figure 2A). Induction of WNT activation after pre-seeding hESCs in our system does not preclude NC induction, as NCs are induced on day 6 following a 2-day pulse with CHIR addition one day after hESCs were pre-seeded in our basal induction media (Figure 2B).

Figure 2. Pre-culture conditions affect whether WNTs require TGFβ signaling to induce NCs.

(A) Schematic of experiments in panels B and C. Note that Day 6 samples in panels B and C were pre-cultured in either hNCi media or mTeSR1 (B) Representative immunofluorescence images on days 5 or 6 after a 2-day pulse with DMSO (CON), or 3μM CHIR (CHIR) as indicated in panel A. SOX10 is Green; PAX7 is red; DAPI is Blue. (n = 3). (C) Representative immunofluorescence images on days 5 or 6 after a 2-day pulse with DMSO (CON), 3μM CHIR (CHIR), or 3μM CHIR+10μM SB43154, as indicated in panel A. SOX10 is Green; PAX7 is red; DAPI is Blue. (n = 3).

Then we tested whether pre-seeding hESCs in mTeSR1 before WNT activation would lead to changes in signaling requirement for hNCi. Freshly dissociated hES were plated for 1 day in mTeSR1 prior to WNT, or WNT+SB treatment. While freshly dissociated hESCs treated with CHIR at seeding produced robust NCs, inclusion of SB, as before, leads to a dramatic reduction in hNCi (Figure 2C). Instead, in hESCs pre-seeded in mTeSR1 for 1 day, then switched to our NC culture conditions, CHIR alone failed to produce hNCs, while CHIR+SB does produce NCs (Figure 2C). These results indicate that pre-seeding in mTeSR1 alters the signaling conditions in hESCs such that WNT activation alone cannot induce NCs, and instead, TGFβ activity must be reduced to enable NC induction upon WNT activation.

To better assess the effect of TGFβ signaling levels on our model of hNCi, we explored hNCi under TGFβ level modulation through combinations of TGFβ and/or SB and quantified SOX10 and PAX7 expression. Robust PAX7 and SOX10 expression was found with our normal induction conditions (Figure 3A). However, the addition of 1ng/mL human recombinant TGFβ1 (hr-TGFβ1) in the presence of CHIR strongly reduced NCs. Similarly, NCi was abrogated when hES were treated with CHIR and SB (1 or 10μM). However, the combination of CHIR plus hr-TGFβ1 and different concentrations of SB revealed robust induction of NC at moderate TGFβ levels. CHIR plus hr-TGFβ1 at low (0.01μM) or high (10μM) SB resulted in low NC induction rates, while CHIR plus 1ng/ml hr-TGFβ1 with intermediate levels of SB (0.1 and 1μM) displayed robust NC induction (Figure 3A). These results indicate that a moderate level of TGFβ signaling is conducive to NC induction, while increased or reduced levels reorient hESC trajectories away from the NC fate.

Figure 3. An optimal level of TGFβ signaling is required for hNCi.

(A) Quantification of SOX10 and PAX7 immunofluorescence images for hNCi on day 5 following a 2-day pulse with 3μM CHIR, 1ng/mL human recombinant TGFβ-1 ligand, and/or different concentrations of SB431542 (SB). (n=3) (B) Representative western blot images of phospho-SMAD2 (pSMAD2), total SMAD2 (SMAD2), or B-ACTIN from hESCs colonies (ES) or hESC colonies pre-treated with SB for 3 hours (ES+SB), or hESCs collected 24 hours after dissociation and treatment with DMSO (CON), 3μM CHIR, or 3μM CHIR+10μM SB (n=3). (C) Quantification of p-SMAD2 relative to SMAD2. Statistics analyzed in panel A by One-way ANOVA followed by Tukey’s post-hoc test, and in panel C by unpaired two-tailed Student’s t-test. * Indicates P<0.05 in panels A and C, and error bars are +/− SEM.

TGFβ activation is mediated by phosphorylation of receptor-regulated SMADs (Massague 2012; Nakao 1997). To test whether our model results in intrinsically moderate TGFβ levels upon which WNT activation leads to NCi, we monitored the presence of SMAD2 and its activated form, pSMAD2. Given that we found sensitivity to SB during the early facets of our culture conditions, we performed western blots for SMAD2 and pSMAD2 on day 1 in cultures treated with DMSO (CON), 3μM CHIR, and 3μM CHIR+10μM SB. As expected, the addition of 10μM SB to hESCs in mTeSR1 for 3 hours before harvest significantly reduced pSMAD2 activity compared with untreated hESC colonies (Figure 3B). Comparatively, TGFβ activity was reduced by ~2.5 fold (250%) in adherent cells treated with DMSO (control) and 3μM CHIR after 1-day post-dissociation. Also, CHIR+SB-treated cells showed significantly reduced pSMAD2 activity relative to cells only treated with CHIR (Figure 3C). These results align with the notion that TGFβ signaling activity is moderate in hESCs following dissociation and suggest TGFβ may be required for NC induction in humans.

Given the limitations to advance our study in early human embryos, we turned to the chick as an avian model organism to investigate the role of TGFβ signaling during NC formation. Whole-mount in situ hybridization of epiblast, Eyal-Giladi XII (EG XII) stage embryos revealed abundant transcription of Tgfbr1, Smad3, and light expression of Smad2 (Figure 4). All three genes were expressed during gastrulation, at Hamburger-Hamilton 4 (HH4) in prospective NCs, and during organogenesis (HH10). Moreover, pSMAD2 activity reflective of TGFβ signaling is known to be active throughout the epiblast in chick embryos during gastrulation stages at HH3-4, in a spatiotemporal profile wherein NCs are specified (Shin et al., 2011b). Our independent immunofluorescence assessment of TGFβ-relevant SMADs and pSMADs concur with published work (Figure 4B). Therefore, signaling components related to TGFβ signaling are localized and functional in prospective NCs.

Figure 4. TGFβ signaling is expressed during early stages of gastrulation in chicken embryos.

(A) In situ hybridization of Smad2 (top row), Smad3 (middle row), and Tgfbr1 (bottom row) in chicken embryos at Eyal-Giladi stage XII, Hamburger-Hamilton (HH)-4 or HH10. Positive-stained cells are lavender. (n=3/stage). (B) Immunofluorescence stained embryos at HH4 with No Primary, SMAD2, P-SMAD2, SMAD3, or P-SMAD3. Top row shows the green channel representing the protein stained and merged with DAPI in blue. The middle row shows the green channel in grayscale, and the bottom row shows the DAPI channel in grayscale. Insets are digital zoom showing punctate P-SMAD expression. (n=3 embryos/ pSMAD stained).

To assess if TGFβ signaling is required during NC formation, we embraced a loss-of-function approach by electroporation of dominant negative (DN-SMAD) plasmid constructs previously shown to specifically target their respective targets (Hata et al 1997; Kretzschmar et al 1999). Chick embryos at stage HH4 were electroporated with an empty vector control plasmid (pCIG) or a mixture of DN-SMAD2/3 plasmids. The contralateral side was left as an untreated internal control and cultured overnight, after which NC formation was established through Pax7 expression (Figure 5A). Normal expression of Pax7 (Basch et al., 2006) in both neural folds was found in 11 embryos that received the control pCIG plasmid on the right side (0/11 affected). Instead, a clear significant reduction of Pax7 expression was detected in 19/22 embryos electroporated unilaterally with the DN-SMAD constructs (P<0.0001, Fisher’s exact test, Figure 5B). A similar reduction in expression of Snai2 was observed in 6/6 embryos electroporated with unilateral DN-SMAD2/3 constructs (Supplementary Figure 2). To further assess the role of TGFβ signaling in vivo in the chick embryo, we adopted an alternative approach to the DN-SMAD electroporations by grafting microbeads pre-soaked in 1 or 10 μM SB431542 (or without SB), unilaterally into the anterior epiblast in prospective NC territory of HH stage 3-4 embryos and incubating them overnight until stages HH 8-10, when they were analyzed for NC markers Pax7 and Snai2 (Figure 5C-D). Expression of both markers was unchanged in treated and untreated neural folds in all 14 embryos incubated with control beads (0/14 affected). By contrast, a significant reduction of NC markers was noted in neural fold territories adjacent of grafted beads pre-soaked with SB at 10 μM (8/8 not shown), and at 1 μM (11/13) doses (P<0.0001, Fisher’s exact test). Sections of embryos with control beads display Pax7 and Snai2 expression of NC markers in treated and untreated neural folds in similar fashion, while sections of embryos receiving the 1 μM SB-soaked beads display reduced Pax7 and Snai2 in the neural fold near the grafted bead (Figure 5D).

Figure 5. TGFβ signaling is required for NC formation in chicken embryos.

(A) Schematic of electroporation experiment. Note the right side of embryos were targeted. (B) Representative dorsal views of immunofluorescence (IF-stained embryos at ~HH8 after electroporation with EGFP-traced electroporated constructs (Green) overlapped with Pax7 (Red). Transverse sections of anterior-posterior areas, as indicated by dashed lines, were counterstained with DAPI (Blue). (pCIG n=11; pCIG-SMAD2,3 DN, n=19/22 electroporated embryos display Pax7 reduction. (C) Graphical representation of the categorical differences observed between groups by IF. (P<0.0001, Fisher’s exact test). (D) Schematic of small-molecule function blocking bead experiment. Note that the left side of embryos were targeted. (E). Representative immunofluorescence stained sections of embryos with beads placed on the left side of embryo at regions indicated in the schematic (a, top row; b, bottom row). Yellow arrows are pointing to expression of Snai2 (Green) and Pax7 (red) in embryos treated with control beds at the treated side; Immunofluorescence counterstained with DAPI (Blue). White arrows point to the same region in embryos incubated with 1μM SB431542. (Snai2 and Pax7 reduced by control beads: 0/14; instead SB431 beads reduced Snai2 and Pax7 on 11/13. (F) Graphical representation of the categorical differences observed between groups by IF.

These results strongly suggest that in chick embryos, TGFβ signaling is required during NC formation, and agree with our human model of NCi, where TGFβ inhibition with SB leads to abrogation of NC formation, further supporting the notion that a moderate level of TGFβ activity is required for neural crest formation in birds and humans alike.

Discussion

Our study found that pre-culture conditions, particularly the use of mTeSR1 maintenance media, could alter the signaling requirements for neural crest induction. When hESC colonies maintained in mTeSR1 were switched to hNCi conditions, WNT activation alone failed to induce neural crest formation. Instead, the addition of TGFβ inhibition was necessary for successful induction (Funa, 2015; Fukuta, 2014). These findings align with previous reports suggesting that the signaling landscape is modulated by the culture environment and that the mesodermal fate may be favored in pre-cultured hESCs when treated with WNT alone (Funa, 2015). The mesodermal markers Brachyury and Goosecoid were expressed in response to WNT activation in hESC colonies maintained in mTeSR1, further emphasizing the shift in lineage potential depending on the culture history of the cells (Shin et al., 2011a).

The role of TGFβ signaling in hNCi was further investigated through a detailed modulation of TGFβ1 levels. Low (0.01μM) or high (10μM) concentrations of SB combined with TGFβ1 resulted in diminished neural crest induction, highlighting the delicate balance of TGFβ signaling required for optimal neural crest differentiation (Leung et al., 2016). These results underscore the necessity for a moderate level of TGFβ activity to maintain the appropriate cellular environment for neural crest cell induction. The response of hESCs to TGFβ modulation further corroborates the importance of TGFβ in guiding the differentiation of pluripotent stem cells into lineage-specific progenitors (Turner et al., 2014). To validate the relevance of these findings in a developmental context, we turned to the chick embryo model, where the positive role of the Smad2/3 mediated TGFβ signaling pathway was shown to be active during gastrulation in the early stages of chick development (Shin et al., 2011b). Our loss-of-function experiments using electroporated DN-SMAD constructs confirmed that TGFβ signaling is essential for proper neural crest induction in vivo.

While this manuscript was in preparation, Rothstein et al 2025 published elegant studies showing a critical role for TGFβ signaling during early NC development. Their findings using a slight variation of our model of human NC induction suggest that TGFβ signaling modulates the axial identity of NC, with TGFβ increasing the cranial identity of hNC induced with low WNT. Our experiments reveal that a moderate level of TGFβ signaling is essential for NC formation under our culture conditions, such that increased or decreased levels are detrimental to NC formation (in agreement with Rothstein et al., 2025). However, we did not assess the impact on axial identity, as all markers we examined indicate a significant reduction or absence of NC formation upon TGFβ signaling modulations tested here. In addition, Rothstein et al show that downregulating the TGFβ1 ligand with morpholins blocks NC formation. Our work is consistent with these results, as we found that combined knockdown of SMAD2 and SMAD3 (via dominant-negative constructs) in HH4 cells resulted in clear downregulation of Pax7 and Snai2 expression in NCs (Figure 5, Supplementary Figure 2). These results complement the findings reported by Rothstein et al and expose a contribution of TGFβ activity in early stages of NC formation in vivo.

Conclusions

In conclusion, our study provides important insights into the role of TGFβ signaling in neural crest induction from human pluripotent stem cells. By demonstrating the critical need for moderate TGFβ activity, we highlight the necessity of fine-tuning signaling pathways to direct hESC differentiation into neural crest progenitors. Furthermore, our findings reinforce the idea that pre-culture conditions and seeding density can influence stem cell sensitivity to TGFβ modulation, making it imperative to optimize experimental conditions to achieve consistent and reproducible neural crest differentiation. This study contributes to our understanding of the molecular regulation of hNCi and offers valuable perspectives for advancing stem cell-based models of human development and disease.

Data Availability:

The data supporting this study's findings are available to the corresponding author upon reasonable request.